Garuda - Garba Rujukan Digital

The Mitigating Overfitting in Sentiment Analysis Insights from CNN-LSTM Hybrid Models Susandri Susandri; Ahmad Zamsuri; Nurliana Nasution; Yoyon Efendi; Hiba Basim Alwan
MATRIK : Jurnal Manajemen, Teknik Informatika dan Rekayasa Komputer Vol. 24 No. 2 (2025)
Publisher : LPPM Universitas Bumigora

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30812/matrik.v24i2.4742

This study aims to improve sentiment analysis accuracy and address overfitting challenges in deep learning models by developing a hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks. The research methodology involved multiple stages, starting with preprocessing a dataset of 5,456 rows. This process included removing duplicate data, empty entries, and neutral sentiments, resulting in 2,685 usable rows. To overcome data quantity limitations, data augmentation expanded the training dataset from 2,148 to 10,740 samples. Data transformation was carried out using tokenization, padding, and embedding techniques, leveraging Word2Vec and GloVe to produce numerical representations of textual data. The hybrid model demonstrated strong performance, achieving a training accuracy of 99.51%, validation accuracy of 99.25%, and testing accuracy of 87.34%, with a loss value of 0.56. Evaluation metrics showed precision, recall, and F1-Score values of 86%, 87%, and 86%, respectively. The hybrid model outperformed individual models, including Convolutional Neural Networks (70% accuracy) and Long Short-Term Memory Networks (81% accuracy). It also surpassed other hybrid models, such as the multiscale Convolutional Neural Network-Long Short-Term Memory Network, which achieved a maximum accuracy of 89.25%. The implications of this study demonstrate that the hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks effectively improves sentiment analysis accuracy while reducing the risk of overfitting, particularly in small or imbalanced datasets. Future research is recommended to enhance data quality, adopt more advanced embedding techniques, and optimize model configurations to achieve better performance.

PELATIHAN DESAIN MEDIA DENGAN APLIKASI CANVA UNTUK SISWA MADRASAH ALIYAH MA'ARIF NU PEKANBARU Zamsuri, Ahmad; Susandri, Susandri; Pane, Eddissyah Putra; Feldiansyah, Feldiansyah; Fajrizal, Fajrizal; Herlina, Sari
Jurnal Pemberdayaan Sosial dan Teknologi Masyarakat Vol 5, No 1 (2025): April 2025
Publisher : Smart Education

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.54314/jpstm.v5i1.3022

Abstract: The Canva application is an application that can be used in creating banner designs. With the utilization and development of techniques in making banner designs. In this case, the use of attractive and appropriate media in the process of making training banners will have a positive impact that will have the potential to greatly improve students' ability to create banners with the Canva application. Students and students really need knowledge in making media design with the aim of providing provisions in making media design even though students at Madrasah Aliyah Ma'arif NU Pekanbaru are quite many who want trainings to improve students' soft skill knowledge. Community Service offers a solution in the form of media design making training with the Canva application for students of Madrasah Aliyah Ma'arif NU Pekanbaru. Where this program is expected to be a forum and inspiration for students to increase their knowledge in making Media Design with the canva application. Where this program is expected to be able to be a forum and inspiration for students to be able to increase their knowledge in making Media Design with the great canva application that students and students of Madrasah Aliyah Ma'arif NU Pekanbaru hope for. As a speaker, he provided socialization and at the same time assistance on how to make Media Design with an interesting and certainly very useful application for students of Madrasah Aliyah Ma'arif NU Pekanbaru. Keyword: Media Design, Canva, Students of Madrasah Aliyah Ma'arif NU Pekanbaru Abstrak: Aplikasi canva merupakan aplikasi yang dapat digunakan dalam pembuatan desain spanduk. Dengan pemanfaatan dan pengembangkan teknik dalam pembuatan desain spanduk. Dalam hal ini, pemanfaatan media yang menarik dan tepat guna dalam proses pembuatan spanduk pelatihan akan memiliki dampak positif yang akan sangat berpotensi meningkatkan kemampuan siswa dalam membuat spanduk dengan aplikasi canva. Para siswa dan siswi sangat membutuhkan pengetahuan dalam membuat desain media dengan tujuan memberikan bekal dalam pembuatan desain media walaupun siswa di Madrasah Aliyah Ma'arif NU Pekanbaru cukup bayak yang menginkan pelatihan-pelatihan untuk meningkatkan pengetahuan softskill siswa. Pengabdian Kepada Masyarakat menawarkan solusi berupa pelatihan pembuatan desain media dengan aplikasi canva untuk siswa Madrasah Aliyah Ma'arif NU Pekanbaru. Di mana program ini diharapkan mampu menjadi wadah dan inspirasi bagi para siswa dapat menambah pengetahuan dalam membuat Desain Media dengan aplikasi canva besar harapan siswa dan siswa Madrasah Aliyah Ma'arif NU Pekanbaru. Sebagai pemateri memberikan sosialisasi dan sekaligus pendampingan mengenai bagaimana pembuatan Desain Media dengan aplikasi canva yang menarik dan pastinya sagat dapat bermanfaat bagi siswa Madrasah Aliyah Ma'arif NU Pekanbaru. Kata kunci: Desain Media, Canva, Siswa Madrasah Aliyah Ma'arif NU Pekanbaru

Enhancing Dental Image Segmentation Techniques: Edge Detection and Color Thresholding Susandri, Susandri; Sumijan , Sumijan; Zamsuri , Ahmad; Rahmiati, Rahmiati; Asparizal, Asparizal
Digital Zone: Jurnal Teknologi Informasi dan Komunikasi Vol. 15 No. 1 (2024): Digital Zone: Jurnal Teknologi Informasi dan Komunikasi
Publisher : Publisher: Fakultas Ilmu Komputer, Institution: Universitas Lancang Kuning

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31849/digitalzone.v15i1.18757

Rapid advancements in medical technology, particularly in the field of dentistry, have led to significant progress in the application of medical imaging techniques to generate valuable image data. The resulting images often exhibit heterogeneous intensity distributions, with boundaries not always distinctly clear between the tooth roots and bone, along with variations in shape and pose. This study specifically aimed to identify the optimal image for segmenting specific parts of the dental structures. Image segmentation is crucial for ensuring effective diagnosis in the context of dental medicine. To achieve optimal dental image segmentation, this research combines edge detection methods with the determination of color thresholds, specifically grayscale and Hue, Saturation, Value (HSV). The research findings revealed that edge detection using the Sobel gradient operator yielded a relevant count of 17,099 pixels. Using RGB=3 and HSV=0.3 the color thresholds show an enhancement in the brightness of the resulting HSV-segmented image, while in the RGB-segmented image, the selected object appears more prominent. The findings of this study contribute significantly to the evolution of dental image segmentation techniques, potentially enhancing the accuracy and effectiveness of diagnoses within the realm of modern dental practice

OPTIMALISASI KINERJA KLASIFIKASI TEKS BERDASARKAN ANALISIS BERBASIS ASPEK DAN MODEL HYBRID DEEP LEARING Salsabila Rabbani; Agustin; Susandri; Rahmiati; M. Khairul Anam
The Indonesian Journal of Computer Science Vol. 13 No. 3 (2024): The Indonesian Journal of Computer Science (IJCS)
Publisher : AI Society & STMIK Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33022/ijcs.v13i3.4034

The conflict between Palestine and Israel has generated strong debates and reactions on social media, including in Indonesia. Public perception of various aspects is certainly important to identify issues in the Palestinian-Israeli conflict. However, the process of manually classifying aspects of the Palestinian-Israeli conflict requires human resources and considerable time. This research aims to explore the views of Indonesians on the Palestinian-Israeli conflict through sentiment analysis based on aspects of Territory, Religion, Politics, and History. Using deep learning technology, specifically a combination model of Convolutional Neural Networks with Long Short-Term Memory (CNN-LSTM), this research analyzes opinion and views data collected from X social media platform (Twitter). This research shows the results of the dataset obtained that the Political aspect dominates more than other aspects. The model evaluation results obtained an accuracy value of 96%, which indicates that the model's ability to classify X users' sentiments towards the Palestinian-Israeli conflict achieved a high level of success.

The Mitigating Overfitting in Sentiment Analysis Insights from CNN-LSTM Hybrid Models Susandri, Susandri; Zamsuri, Ahmad; Nasution, Nurliana; Efendi, Yoyon; Alwan, Hiba Basim
MATRIK : Jurnal Manajemen, Teknik Informatika dan Rekayasa Komputer Vol. 24 No. 2 (2025)
Publisher : Universitas Bumigora

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30812/matrik.v24i2.4742

This study aims to improve sentiment analysis accuracy and address overfitting challenges in deep learning models by developing a hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks. The research methodology involved multiple stages, starting with preprocessing a dataset of 5,456 rows. This process included removing duplicate data, empty entries, and neutral sentiments, resulting in 2,685 usable rows. To overcome data quantity limitations, data augmentation expanded the training dataset from 2,148 to 10,740 samples. Data transformation was carried out using tokenization, padding, and embedding techniques, leveraging Word2Vec and GloVe to produce numerical representations of textual data. The hybrid model demonstrated strong performance, achieving a training accuracy of 99.51%, validation accuracy of 99.25%, and testing accuracy of 87.34%, with a loss value of 0.56. Evaluation metrics showed precision, recall, and F1-Score values of 86%, 87%, and 86%, respectively. The hybrid model outperformed individual models, including Convolutional Neural Networks (70% accuracy) and Long Short-Term Memory Networks (81% accuracy). It also surpassed other hybrid models, such as the multiscale Convolutional Neural Network-Long Short-Term Memory Network, which achieved a maximum accuracy of 89.25%. The implications of this study demonstrate that the hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks effectively improves sentiment analysis accuracy while reducing the risk of overfitting, particularly in small or imbalanced datasets. Future research is recommended to enhance data quality, adopt more advanced embedding techniques, and optimize model configurations to achieve better performance.

Spatial-Temporal Analysis of Earthquakes in Indonesia with Deep Learning Models: Performance Evaluation of CNN, LSTM, and Hybrid CNN-GRU Susandri, Susandri; Fajrizal, Fajrizal; Bakri Nasution, Feldiansyah
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 9 No 5 (2025): October 2025
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v9i5.6538

Indonesia, located along the Pacific Ring of Fire, experiences high seismic activity with over 6,000 earthquakes annually. Accurate earthquake prediction remains a major challenge because of the complexity of geological dynamics and limitations of traditional methods in capturing nonlinear seismic patterns. Although deep learning approaches have shown promise, previous studies have often treated spatial and temporal analyses separately, limiting holistic predictive performance. This study proposes a novel hybrid CNN-GRU deep learning model that integrates spatial feature extraction CNN and temporal sequence modeling GRU, and compares its performance with of that CNN, LSTM, GRU, and Bidirectional LSTM using a dataset of 117,251 earthquake events in Indonesia (2008–2024). The results show that Bidirectional LSTM achieved the best temporal accuracy (R² 0.653, RMSE 0.592), while the hybrid CNN-GRU provided balanced spatial-temporal performance (R² 0.587). Notably, the performance gap between Bidirectional LSTM and other models (e.g., Hybrid CNN-GRU) was statistically validated via paired t-test (p < 0.05). The proposed models generalize well to unseen regions such as Maluku-Papua. The key contribution is the hybridization of spatial-temporal learning in a single-model architecture - where CNN processes latitude-longitude coordinates via 1D convolutions while GRU handles temporal sequences - an approach lacking in earlier works. This directly improves early warning systems in seismically active areas by providing 32% higher accuracy than conventional methods.

The Optimizing Sales Strategies to Address Excessive Stock Accumulation: A Data Mining Approach Susandri; Muhammad Arief Solihin; Hamdani; Asparizal
JAIA - Journal of Artificial Intelligence and Applications Vol. 4 No. 1 (2024): JAIA - Journal of Artificial Intelligence and Applications
Publisher : STMIK Amik Riau

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33372/jaia.v4i1.1110

The Two Pelita Weaving Business has recorded significant sales in the weaving industry, despite facing challenges in managing product stock due to the accumulation of excess stock caused by a lack of customer interest. This study employs data mining techniques, specifically the Association Rule and Apriori algorithms, to analyze sales patterns. The analysis results using Python and Orange Data Mining showed consistency in the relationship between Siku Keluang Weaving and Pucuk Rebung Weaving products, with high occurrence rates of purchase patterns (11.74% and 10%, respectively). High confidence levels with Python at 96.36% and Orange Data Mining at 99.1% indicate that customers who purchase Siku Keluang Weaving are also likely to purchase Pucuk Rebung Weaving products.

Sentimen Pengguna Aplikasi BRImo: Kinerja Algoritma Support Vector Machine, Naive Bayes, dan Adaboost Susandri; Yurnalis; Edwar Ali; Susanti; Asparizal
SATIN - Sains dan Teknologi Informasi Vol 9 No 2 (2023): SATIN - Sains dan Teknologi Informasi
Publisher : STMIK Amik Riau

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33372/stn.v9i2.1057

Dalam konteks perkembangan industri perbankan yang semakin maju, pemanfaatan teknologi modern menjadi faktor kunci untuk meningkatkan kualitas layanan dan memenangkan persaingan di era digital. Bank Rakyat Indonesia (BRI) memikat perhatian masyarakat melalui peluncuran aplikasi perbankan seluler, BRImo. Namun Bank ini perlu meraih pandangan dan pengalaman nasabah terhadap aplikasi mobile banking untuk meningkatkan kualitas pelayanan. Penelitian ini memiliki tujuan untuk menganalisis ulasan pengguna BRImo sebagai objek penelitian. Komparasi dilakukan antara algoritma Support Vector Machine (SVM), Naive Bayes (NB), dan Adaboost dalam mengolah data teks. Evaluasi dilakukan berdasarkan tingkat akurasi, presisi, recall, dan nilai F1-score. Hasil penelitian menunjukkan bahwa algoritma SVM memberikan kinerja terbaik dalam mengklasifikasikan tanggapan masyarakat terhadap aplikasi BRImo, dengan tingkat akurasi sebesar 90,4%, presisi 90,8%, recall 90%, dan nilai F1-score 90,3%. Sebagai perbandingan, algoritma Adaboost memberikan nilai terendah dengan tingkat akurasi sebesar 87%, presisi 87,2%, recall 86,8%, dan nilai F1-score 86,9%.

Evaluation of the Effect Of Regularization on Neural Networks for Regression Prediction: A Case Study of MLLP, CNN, and FNN Models Susandri; Zamsuri, Ahmad; Nasution, Nurliana; Ramadhani, Maya
INOVTEK Polbeng - Seri Informatika Vol. 10 No. 3 (2025): November
Publisher : P3M Politeknik Negeri Bengkalis

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35314/m2rcsf96

Regularization is an important technique for developing deep learning models to improve generalization and reduce overfitting. This study evaluated the effect of regularization on the performance of neural network models in regression prediction tasks using earthquake data. We compare Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Feedforward Neural Network (FNN) architectures with L2 and Dropout regularization. The experimental results show that MLP without regularization achieved the best performance (RMSE: 0.500, MAE: 0.380, R²: 0.625), although prone to overfitting. CNN performed poorly on tabular data, while FNN showed marginal improvement with deeper layers. The novelty of this study lies in a comparative evaluation of regularization strategies across multiple architectures for earthquake regression prediction, highlighting practical implications for early warning systems.

ANALISIS SENTIMEN PERUNDUNGAN TERHADAP GURU DENGAN MENGGUNAKAN METODE SUPPORT VECTOR MACHINE DAN NAÏVE BAYES Zamzuri, Ahmad; Nasution, Nurliana; Susandri, Susandri; Bimby, Novia Putri
JOURNAL OF SCIENCE AND SOCIAL RESEARCH Vol 8, No 4 (2025): November 2025
Publisher : Smart Education

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.54314/jssr.v8i4.4916

Abstract: This study discusses sentiment analysis of bullying experienced by teachers on social media. The research employs the Support vector machine (SVM) and Naïve Bayes methods to classify sentiments into positive, negative, or neutral categories. The data were collected from various social media platforms and analyzed using text mining techniques. The results show that the SVM method achieved a higher accuracy rate compared to Naïve Bayes in detecting negative sentiments related to bullying toward teachers. These findings contribute to a better understanding of digital bullying patterns targeting educators and provide a foundation for developing more effective policies to address bullying cases in the educational environment. Keyword: sentiment analysis, bullying, teachers, support vector machine, naïve bayes, text mining. Abstrak: Penelitian ini membahas analisis sentimen terhadap perundungan yang dialami oleh guru di media sosial. penelitian ini menggunakan metode support vector machine (svm) dan naïve bayes untuk mengklasifikasikan sentimen menjadi positif, negatif, atau netral. data yang digunakan berasal dari berbagai platform media sosial dan dianalisis menggunakan teknik text mining. hasil penelitian menunjukkan bahwa metode svm memiliki tingkat akurasi yang lebih tinggi dibandingkan dengan naïve bayes dalam mendeteksi sentimen negatif terkait perundungan terhadap guru. temuan ini dapat membantu dalam memahami pola perundungan digital terhadap tenaga pendidik serta memberikan dasar untuk kebijakan yang lebih efektif dalam menangani kasus perundungan di dunia pendidikan. Kata kunci: analisis sentimen, perundungan, guru, support vector machine, naïve bayes, text mining..

Title

Found 40 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 40 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 40 Documents
Search